A new low.

[Spoonwood]

which I don't do, and advise others against, since building scouts and collecting goody huts is a crutch for lousy trading skills, only useful for the first quarter of the game, and still unreliable even then.

You can also pop settlers or cities. Honestly, that's the trait. There exists no "crutch" here, as there exists no growth crutch by deciding to play agricultural, or no production or commerce crutch by deciding to play industrious, or no corruption crutch by playing commercial, or no research crutch by playing scientific, or no veteran and elite unit crutch by playing militaristic. Also, last time I checked a slew of fast research games have gotten played with Russia (including ones at Emperor level)... see here http://hof.civfanatics.net/civ3/index.php?condition=Diplomatic&mapSize=Any&submit=Go and here
http://hof.civfanatics.net/civ3/index.php?condition=Spaceship&mapSize=Any&submit=Go

Even science civs tend to go for Feudalism first. It definitely can matter what tech you research after Banking and Chemistry, no matter the level, if the AIs can outresearch the human player. You don't know what type of science commerce Aegis produces in general, nor on the map she will play.

 

[Sweetchuck]

Question - did you fortify that musketman and upgrade your whole lot from another location?

 

[darski]

It was interesting to happen upon this thread just coming off a game as Japan. My advisor told me that the best unit the French had was a warrior so I decided it was time to take some more land.

I sent 5 cavs and 4 Samurai to Paris. One ***Reg*** spear killed 2 cavs, 2 sams and redlined another cav before the 4th cav finally took him out. when y'all say that this can be explained by math... I am not prepared to believe you any more. the game is weighted against the human player

... I ain't going to believe anything else.

 

[Spoonwood]

Your beliefs matter little to how the game plays out Darski. Did Paris lie on a hill? What size was it? Did it promote at all after the first battles? The spear might not have had a defense of 2 with the bonuses Paris had. So, losing 2 samurias might actually come out as expected. Cavs losing or retreating also does happen. You've given us only a few instances of improbable behavior at best. So, that's very little to make a rational argument that the game gets weighted against the human player. I can easily give a few instances of where I've popped an SGL on the first tech I've researched and then conclude the game biased in my favor from that little information.

 

[Padma]

It's all mathematics and probability (which comes back to mathematics ).

The biggest problem is most people don't really grasp what random numbers are really like. A truly random set of numbers can easily generate a series of terrible losses for the player. And cognitive bias rears its ugly head, when we don't give equal credence to the times we inflict just as unlikely losses to the AI.

 

[capnvonbaron]

You can also pop settlers or cities. Honestly, that's the trait. There exists no "crutch" here, as there exists no growth crutch by deciding to play agricultural, or no production or commerce crutch by deciding to play industrious, or no corruption crutch by playing commercial, or no research crutch by playing scientific, or no veteran and elite unit crutch by playing militaristic. Also, last time I checked a slew of fast research games have gotten played with Russia (including ones at Emperor level)

Every one of the other traits require some game play to use correctly though. You can't make more science, commerce, agri, or production if you don't build workers, expand intelligently, defend yourself, and manage it all correctly. Expansionist, in every game I've played, is either feast or famine; either you get a lot of goodies, and its worth it, or nothing great comes out of it and you might as well be playing a civ with one trait (and suffer the rest of the game as such). While yes it might benefit a fast research game, I maintain that it requires no skill to gather a bunch of huts and *hope* they have what you want. Anything that requires no skill to use and requires luck to bear any benefit is a either a crutch or a cheap exploit; as a trait, expansionist is a crutch.

Even science civs tend to go for Feudalism first.

I didn't say otherwise. Yes they research feudalism first, but I was pointing out that they get theology free 99% of the time, so its almost always available before you get done with engineering.

And no, it doesn't matter what one researches after chem/banking. There is really only one road to go to get to the IA after that, unless you are talking about a culture victory and want SHake's quickly. Either the AI has a tech lead by then, or not. If the AI has what you want, buy/beat/trade techs out of them. If you've built a halfway decent empire by this point, you should be outresearching all the numbskull AI civs that are busy getting the crap non-required techs. If you have two or three other powerhouse civs and kill their economy by the means I mentioned earlier, then you should be out-researching them. Especially so if its a continents/archipelago map, which I would suggest Aegis use until comfortable with a rapid tech pace.

Aegis isn't trying for a HoF victory, Spoon, he/she is trying for a fun and successful victory at a higher difficulty, and thats what we are supposed to be helping with.

 

[Pyrrhos]

It was interesting to happen upon this thread just coming off a game as Japan. My advisor told me that the best unit the French had was a warrior so I decided it was time to take some more land.

I sent 5 cavs and 4 Samurai to Paris. One ***Reg*** spear killed 2 cavs, 2 sams and redlined another cav before the 4th cav finally took him out. when y'all say that this can be explained by math... I am not prepared to believe you any more. the game is weighted against the human player

... I ain't going to believe anything else.

That's a turn-around, Darski!

There is, however, a cure - a) make certain "preserve random seed" is turned off, + b) save before every combat & reload. Logic: "The PC/game cheats, I cheat..."

 

[capnvonbaron]

I am not prepared to believe you any more. the game is weighted against the human player

Heheh... yeah... you just need to write down and remember the time you had an even-match battle that ended up in a crushing victory. In the same game, I had a veteren Rider take down a vet fortified Musketeer and not lose a single HP, and several battles where the vet riders did the same and lost perhaps 1 hp. Once a Rider was reduced to 1HP against a Musketeer fortified on a hill, and STILL WON. I was amazed. It seemed like the RNG was not in Joan's favor at all... until The Battle For Paris; a reg fortified pikeman took out one vet rider and redlined another before I finally killed him. It took every available Rider I had to take Paris, a force of around 25 against a garrison of 6 in a size 10 city. Thank god most of the losing ones retreated... I've had other games where I swear the damned things forgot they could retreat.

The best thing to do... always make sure you have units in reserve in case the initial attack fails. If you do that, you won't be disappointed. If you don't, expect to be disappointed and be pleased if it works in your favor!

 

[ThinkTank]

It's all mathematics and probability (which comes back to mathematics ).

The biggest problem is most people don't really grasp what random numbers are really like. A truly random set of numbers can easily generate a series of terrible losses for the player. And cognitive bias rears its ugly head, when we don't give equal credence to the times we inflict just as unlikely losses to the AI.

Amen. Easily the best statement on this forum of 2009.

Have mercy on you, o ye of little faith, unbeknownesthththt with the ghhhhospel of Padma. Ye may want to read this:

John Allan Paules - Innumeracy: Mathematical Illiteracy and Its Consequences.

 

[Spoonwood]

Aegis isn't trying for a HoF victory, Spoon, he/she is trying for a fun and successful victory at a higher difficulty, and thats what we are supposed to be helping with.

Hence I recommended she play builder early, and still play for her conquest victory. I think I'd just rehash what I said earlier otherwise with respect to much of what you wrote otherwise... other than that I think them getting *Monotheism* as their free tech 99% of the time only happens in Vanilla, not in Conquests.

Seconds what Padma says.

Bumps what Capnvonbron says... before you decide the game against you, you need a full record of many battles and other RNG events *that have gone in your favor*. A 1 HP rider against a fortified musketman on a hill winning certainly comes out as rather improbable. People have gotten two SGLs in a row... even with a scientific tribe, that has less than 1% probability of happening.

Seconds, thirds, and fourths ThinkTank's recommendation.

 

[capnvonbaron]

... other than that I think them getting *Monotheism* as their free tech 99% of the time only happens in Vanilla, not in Conquests.

Whoops... monotheism. Thats what I meant to say Yeah some confusion might've come from that mis-statement...

 

[Pyrrhos]

Amen. Easily the best statement on this forum of 2009.

Have mercy on you, o ye of little faith, unbeknownesthththt with the ghhhhospel of Padma. Ye may want to read this:

John Allan Paules - Innumeracy: Mathematical Illiteracy and Its Consequences.

Do you yourself belong to the mathematically illiterate? Let's take a look at the maths of Darski's example:

I sent 5 cavs and 4 Samurai to Paris. One ***Reg*** spear killed 2 cavs, 2 sams and redlined another cav before the 4th cav finally took him out.

As a regular spear has three hit points, it cannot have lost more than (two plus promotions to vet & elite) four rounds of combat whilst winning at least 19 rounds of combat (Darski is not an inexperienced player and would have used at least vet units). Since it was Paris, an AI capital, there would be no defensive bonus from terrain. But the spear was fortified and, for arguments sake, let's assume that Paris was a town, which is what it sounds like from the description given by the mil adviser. This gives the spear a total bonus of 50%, or a defensive combat value of 2+(2x50%)=3. Furthermore, let us assume that the four rounds won by Darski's first four units was divided 3-1 in favour of her cavalry over her samurai. This gives us the following maths:

Per round of combat
Cavalry vs Spearman 6/6+3 = 66.7% cav win, 33.3% spearman win
Samurai vs Spearman 4/4+3 = 57.1% samurai win, 42.9% spearman win

The whole sequence until the 4th cavalry attacked:
Vs cavalry, 14 rounds in all 3 of which was won by the cavalry and 11 by the spearman. This gives an overall ratio of 21.4% against 78.6%, a statistical deviation greater than 3!

Vs samurai, 9 rounds in all of which the samurai won 1 and the spearman 8 which gives the ratio 11.1% vs 88.9%, a statistical deviation greater than five!

The mathematical probability of the spearman winning all those rounds works out at 0.000000006424696131 or 1 in 155,649,627. (The probability of winning the lottery is given as a chance of between 1 in 5,200,000 to 1 in 100,000,000.)

Science tells us, provided of course that Darski's recapitulation is factual, that the PRNG of Civ III is not working properly, that is, it does not obey the laws of mathematical probability.

You and Padma would do well to remember the following - even Nobel Laureate level science starts with a hunch or as you like to term it, "cognitive bias"! When you dismiss it they way you do, it says A LOT about yourselves and your cognitive abilities...

 

[Chamnix]

Welcome back, Pyrrhos !

 

[Pyrrhos]

Thx!

Incorrigible as ever, I'm afraid!

 

[Spoonwood]

Pyrrhos,

Paris may have stood on a hill. Darski simply hasn't told us enough information to know how improbable her string comes out as. We don't know the math of her example. Since we don't know the math of her example, her statement really only tells us that she won't *accept* (not "believe" so much) mathematical results if she wants to believe another way. We don't know if she attacked the spear on the same turn either, so it could have healed several times. Cognitive bias itself comes as fine... as long as you remember that cognitive bias exists. Dismissal of it usually just means that someone believes the other person seems to have forgotten it exists. From her comments here and elsewhere (see below) Darski simply doesn't seem aware of it (nor would she necessarily care).

I honestly don't believe this correct "Darski is not an inexperienced player and would have used at least vet units", as even though Darski has played the game for a while, she still hasn't made it much past Regent from what she's written on these forums. She's also said before that if she doesn't get the Republic slingshot she can't get The Republic in less than 50 turns... and has claimed to understand everything about commerce. Mind you, Regent has a fairly low tech cost, and people can get less The Republic in less than 50 turns at Deity level where tech costs much more.

Whatever the actual statitistical probability of her *5* battles, it only comes out as *5* battles (the 6th battle doesn't count here, since she won). This comes as a *very* small sample size in comparison to the number of RNG evens in really any game, almost surely including, though not necesarily limited to, what the AIs research, win/loss of battles, retreat of fast units, SGLs coming out, the AI declaring war, what goodie huts give you. That sounds like easily thousands and thousands of RNG events over the course of many games. So, even granting your assumptions about her game as correct, that "Science tells us, provided of course that Darski's recapitulation is factual, that the PRNG of Civ III is not working properly, that is, it does not obey the laws of mathematical probability." simply does not hold. The "laws" of mathematical probability, apply when the limit *approaches positive infinity*. As an example, that the probability of a die roll equals 6, 1/6 of the time holds for a sample size x, *when x approaches infinity*. In other words, we only have 1/6 as the real probabilty of the die throw *for an infinite sample size*. At best, Darski has a sample size of 5... which comes out as far, far too small to approximate the behavior of an infinite sample size. Standard deviations come from the infinite sample size.

Popping two consecutive SGLs with a scientific tribe comes out as having a probability of .0025. We have a variance of .095 here (.05*2*.95) and an expected value of .1. So we have a standard deviation .3082207 here. So, we have a statistical deviation greater than 6 here (.1+.3082207*6<2)!!!!! By the reasoning you used above, we have an event that defies the laws of mathematical probability! Since I've seen threads where people do that, the game must clearly have a bias in the human's favor.

No... the sample size simply comes out as *far* too small. So, conclusions about bias of the game *merely* reflect cognitive bias.

 

[Jokeslayer]

~lots of nonsense~

See the part where the probability is greater than 0?

Yeah.

 

[vmxa]

And cognitive bias rears its ugly head, when we don't give equal credence to the times we inflict just as unlikely losses to the AI.

So true, this is why top pool players use drills to practice, not playing games. If you play pool by yourself to practice you are likely to leave remembering all good parts and over looking the misses.

 

[vmxa]

You and Padma would do well to remember the following - even Nobel Laureate level science starts with a hunch or as you like to term it, "cognitive bias"! When you dismiss it they way you do, it says A LOT about yourselves and your cognitive abilities...

Not to get involved in the out come, but there exist a term for what happen. It is known as defying the odds. It does not mean the rng is broken. I think it is very flawed, but the needed samples to prove it we are not going to have.

The thing is, I doubt one could convince someone that the game is not rigged, with logic.

It does not even matter if it is broken or not, It is what we have and we have to deal with it or not play.

 

[Pyrrhos]

Pyrrhos,

Paris may have stood on a hill.

No.

In that game, France was an AI civ and Paris is the capital, i.e. the starting location. Just run a sample of world maps in the editor and tell me if you come across a starting location on a hill.

Darski simply hasn't told us enough information

Which I have acknowledged. Furthermore, I've known Darski a long time, and she has never stooped as low as to deliberate falsehood. I accept her word for it, especially as she used to be one of the people who defended the pRNG assiduously. Thats said, she did call me TROLL because I maintained that the combat system was fubar...

The rest of your post, we can leave without comment apart from one thing: You make a claim based on a sample of three, yet derides a sample of 25 as being "*far* too small". Now, that's what I call cognitive bias!

 

[Pyrrhos]

Not to get involved in the out come, but there exist a term for what happen. It is known as defying the odds. It does not mean the rng is broken. I think it is very flawed, but the needed samples to prove it we are not going to have.

The thing is, I doubt one could convince someone that the game is not rigged, with logic.

It does not even matter if it is broken or not, It is what we have and we have to deal with it or not play.

Of course, but since it's not a RNG but a pRNG, it's built into the game - intentionally.

Whilst one cannot have anything but admiration for Sid for the ingenious way in which he designed such a complex game to fit within the limitations of DOS 5 and it's allowance of no more than some 450 to 525 (or so) KB, by 2001 computer power had advanced by a factor of a thousand to the point where it would have been possible to revamp the combat system. That this was not done is not an admission of neglect on part of the CIV III design team, but rather that the mechanics of the combat system fit into the game they wanted to create, a game where it is possible for Stone or Bronze Age technology to "defy the odds" and successfully take on a modern Armoured Division equipped with M1A2, Leopard A4 or Leclerc MBTs.